Generally, the rapid, accurate, sensitive, and specific analysis of microscopic elements such as cells, microbes, and particulates (hereafter, “particles”) is of great importance in cell biology research, pharmaceutical research and development, microbiology, hematological diagnostics, water and milk quality testing, paint and emulsion production, and other applications. Historically, inspection and characterization of particles were carried out on optical microscopes, which are still in widespread use in medical, research, and industrial laboratories worldwide. However, even as microscopes have become increasingly sophisticated, with functions such as, e.g., fluorescence detection, confocal geometry, and superresolution, they continue to be severely limited in analysis speed and sample throughput. Collection of image information at high enough resolution to allow characterization and identification of human cells is certainly possible, but at a steep price in terms of the total number of individual particles that can be so analyzed in any reasonable amount of time.
For the last several decades, an alternative modality of analysis, characterization, and counting of particles has emerged and taken root—flow cytometry. This technology contrasts with microscopy in a couple of key ways: it works by flowing a liquid sample across a single fixed point of interrogation, rather than interrogating (whether by scanning or imaging) over a fixed sample; and it generally analyzes information averaged over each entire particle, rather than resolving details at, e.g., a subcellular or even subnuclear level. The result is a trade-off between spatial resolution and speed: microscopy excels at the former, flow cytometry at the latter. Accordingly, the two techniques have in recent years co-evolved by commanding complementary applications, often performed in the same laboratory and even side by side: For example, routine hematology diagnostics is near-universally carried out on highly efficient automated analyzers built on flow cytometric principles, but time-consuming microscope-based review of blood slides is just as universally performed to clarify by high-resolution visual inspection those instances where the flow cytometer results are inconclusive.
Generally, contemporary flow cytometric analysis relies on several established interrogation techniques to extract identifying information from each passing particle: dc electrical impedance, ac electrical conductivity, optical extinction, light scattering, and fluorescence. Dc impedance is typically very limited in the ability to distinguish cells similar in volume but different in composition; it is therefore generally used only for first-line hematological screening, or as add-on to optical methods on advanced analyzers. Ac electrical conductivity and optical extinction can each add a dimension of differentiation to help distinguish similar cells, but not in isolation. Typically, the workhorses of flow cytometry are (elastic) light scattering and fluorescence.
Generally, scattering depends on refractive index variations either between the particle and its liquid surroundings, or within the particle itself; it gives relatively coarse information on presence, size, and rough morphology of the particle itself and (in the case of a cell) any nucleus and/or cytoplasmic bodies present. Scattering works on intrinsic properties of particles and does not generally require the use of expensive reagents. However, it does not effectively differentiate between similarly sized and structured, but functionally distinct, particles (e.g., a lymphocyte and a nucleated erythrocyte; or two microbeads surface-functionalized with different analyte-specific antibodies).
Fluorescence is by far the optical interrogation method most widely used in flow cytometry. A fluorophore can be intrinsic (such as tryptophan); externally introduced as a standalone agent (such as propidium iodide); or externally introduced as a conjugate to a particular antibody (such as, e.g., phycoerythrin conjugated to CD4). There are further variations, such as tandem dyes and quantum dots, which while useful, may not be directly relevant here. Autofluorescence from compounds naturally or biologically stimulated within a cell are of interest in certain research areas; however, the broadest application of fluorescence in flow cytometry is based on the introduction of external fluorophores. Non-conjugated fluorophores are typically used to bind selectively to major cell constituents such as DNA or RNA, and return information on, e.g., the presence or absence of a nucleus, or on the maturation stage of certain blood cells; but they are not particularly specific. They are therefore used in combination with either scattering (in the context, e.g., of hematological diagnosis) or panels of fluorescent antibody conjugates.
Antibody-conjugated fluorophores are commonly called “tags” or “labels” in that they perform the function of selectively identifying a particle based on the presence of matching antigens on the particle surface; also used are fluorescent labels that bind to and identify intracellular elements, such as, e.g., actin or cytokeratin. As an example of antibody tagging in peripheral blood samples, leukocytes can be selectively identified out of the much vaster population of erythrocytes, in company of which they are found, by incubating the sample with a fluorophore conjugated to the CD45 antibody (which selectively binds to the matching CD45 antigen on the leukocyte surface). Flow cytometric analysis of the incubated sample results in leukocytes, but not erythrocytes, generating fluorescence upon passing through the interrogation region. Tagging by antibody conjugation and fluorescence analysis by flow cytometry has enabled great strides in the field of immunology, where cell type, function, and even stage of development are associated with the expression of distinct sets of surface antigens.
The development and manufacture of fluorescent antibody conjugates—particularly monoclonal antibodies (MAbs)—is, however, cumbersome and expensive. While some recent advances have been made in the generation of synthetic antibodies, most antibodies are produced with the use of animal-derived cells in a lengthy process. And while several long-established fluorescent dyes are relatively inexpensive, the drive in flow cytometry toward greater and greater multiplexing—i.e., the concurrent use of multiple tags, and therefore multiple fluorophores, in the same assay—has pushed the field into a great proliferation of custom, relatively expensive, fluorophore compounds. In addition, the process of conjugation itself is an additional factor in the final cost of the antibody conjugate.
Typical research assays performed on flow cytometers include ones for transfection, cell signaling, cell lineage, and stem cells; many of these assays, especially in immunology, require the concurrent use of cocktails of large numbers of fluorescent MAb conjugates, resulting in relatively high costs even for experiments with few repeats. Clinical practice is, by contrast, generally focused on the use of relatively few labels—for example, in management of AIDS patients, where only two or three separate MAbs are required for each test—but the relative frequency of such tests in a typical laboratory is much greater. In either research or clinical flow cytometry applications, therefore, expensive reagents used in selective labeling of cells or microparticles are a definite concern. Annual reagent expenditures in both research and clinical laboratories—expenditures dominated by fluorescent MAb conjugate labels—can be comparable to, and sometimes exceed, the capital cost of a flow cytometer itself, which typically has a life span of five years or more.
Additionally, there is the issue of fragility. MAbs and MAbs-based labels typically have to be kept under strict refrigeration protocols, and thawed carefully just prior to use. This means operational complexity in terms of both proper storage at or near the point of use (hospital or institution) and proper handling during the entire chain of custody from manufacturer to end user. In developed countries this complexity is manageable but drives up the cost assays using MAb reagents; in developing countries it often means that those assays cannot be performed. In resource-poor countries, power shortages, unreliable distribution networks, and limited means of refrigeration, not to mention more general financial constraints, make use of MAb-conjugated fluorophore labels always challenging and often impossible.
Also, whether dealing with MAb tags or with non-conjugated fluorophores, every assay requiring labels involves at least one, and frequently more, incubation steps. Incubation conditions like temperature and time vary considerably, but it is common for a single surface-antigen MAb incubation step to take 15 minutes. This delay in the workflow is significant for a laboratory performing routine tests, and—together with the high cost of MAbs—has limited widespread adoption of MAb assays for clinical use to those that currently cannot be reliably performed any other way, like the CD4-positive T-lymphocyte test for AIDS patients.
Moreover, the tolerance toward difficult and time-consuming sample preparation steps may be relatively high in the context of a clinical laboratory, and very high in a research laboratory, but it is generally extremely low in industrial, or industrial-scale, process monitoring steps, where throughput-limiting (and costly) bottlenecks can mean the difference between economically viable and nonviable production. Such, for example, is the case of milk quality testing, currently performed mainly in central reference laboratories equipped with high-throughput flow cytometric instruments. Testing for water quality has similar constraints. In both cases both a high throughput and a low cost per assay are operational imperatives.
In the field of microscopy, certain advances in the use of spectroscopy have made it possible to analyze specimens without the need for expensive reagents like MAb-conjugated fluorophore labels. Spectroscopy generally provides detailed information about the intrinsic chemical composition of a sample by interrogating the sample with optical radiation. Absorption spectroscopy in its various forms can be very sensitive, but it is not well suited to water-rich samples and aqueous solutions and suspensions. In such cases, inelastic scattering spectroscopy techniques are more effective. These techniques differ from conventional (elastic) light scattering as commonly used in flow cytometry in that they rely on inelastic light scattering—i.e., based on the transfer of energy from the interrogating photons to the material under analysis or vice versa. Raman scattering, resonant Raman scattering, and Coherent Anti-Stokes Raman Scattering (CARS) are examples of such techniques that have been successfully applied to the study of fixed specimens (as on microscope slides) or, more recently, in vivo with small animals. In CARS, pulses at two different optical frequencies (νpump and νStokes) are combined at the sample; the nonlinear optical interaction that results from the high peak powers (due to the short pulse duration) is responsible for the CARS signal at a third optical frequency (νanti-Stokes=2νpump−νStokes). In particular, variations on the CARS technique, such as forward CARS (F-CARS); epi-CARS; polarized CARS (P-CARS); multiplex CARS (M-CARS); broadband CARS (B-CARS) and closely related approaches, such as Stimulated Raman Scattering (SRS), have all been demonstrated in biomicroscopy applications. Each of these techniques is an improvement over traditional microscopy in that it provides detailed information about the chemical make-up of the sample, greatly adding specificity to the analysis.
However, despite advances that enable real-time video-rate analysis of a specimen, all of these techniques are far too slow, by several orders of magnitude, to compete with flow cytometry in the sheer number of cells or other particles that can be analyzed in a given unit of time. Accordingly, they have overwhelmingly been developed for investigation of stationary samples, and are therefore unsuitable for the kind of ensemble-wide cellular assays provided by flow cytometry.
There have been attempts at combining the superior chemical specificity of spectroscopy with the sample interrogation framework of flow cytometry. For example, experimental results in recent years demonstrated the application of particular versions of CARS spectroscopy to samples flowing in microchannels. However, these results suffered from severe limitations. In the study by Wang et al. [Wang et al., Optics Express 16, 5782 (2008)], schematically represented in FIG. 1(a), an apparatus 100 includes a mode-locked laser oscillator 110 and a mode-locked laser oscillator 120, a synchronization control module 130, a flowcell 140 containing a flow channel, and a microscope objective 102. Mode-locked laser oscillators 110 and 120 are synchronized by a control signal 101 produced by synchronization control module 130; mode-locked laser oscillator 110 generates a pump pulse train 103 at a wavelength corresponding to the desired CARS pump wavelength, and mode-locked laser oscillator 120 generates a Stokes pulse train 105 at a wavelength corresponding to the desired CARS Stokes wavelength; the pump and Stokes pulse trains are combined and focused by objective 102 onto particles for analysis in the flow channel of flowcell 140. Pulse trains 103 and 105 are both a series of pulses which interact with a particle such that multiple pulses in the train interact with the particle. None of these pulses in these trains are chirped.
In this example, the CARS analysis is restricted to a single vibrational CARS (anti-Stokes) frequency corresponding to the difference between the CARS pump frequency of pump pulse train 103 and the CARS Stokes frequency of Stokes pulse train 105. Additionally, this apparatus achieves a sample flow velocity of about 40 mm/s—almost three orders of magnitude slower than a typical commercial flow cytometer and unsuitable for laboratory-based analysis of real samples in a reasonable time.
In another example, the study by Camp Jr. et al. [Camp Jr. et al., Optics Express 17, 22879 (2009)] schematically represented in FIG. 1(b), an apparatus 150 includes a mode-locked laser oscillator 160, a photonic crystal fiber 170, a flowcell 180 containing a flow channel, and a microscope objective 152. Mode-locked laser oscillator 160 generates a pulse train 153 at a wavelength corresponding to the desired CARS pump wavelength; a portion of such pulse train is split and fed into photonic crystal fiber 170, while the rest continues as a pump pulse train 155. Photonic crystal fiber 170 generates by nonlinear optical processes a Stokes pulse train 157 with light spread across a band of wavelengths, each acting as a CARS Stokes wavelength; the pump and Stokes pulse trains are combined and focused by objective 152 onto particles for analysis in the flow channel of flowcell 180. Pulse trains 153 and 155 are both a series of pulses that interact with a particle such that multiple pulses in the train interact with the particle. None of these pulses in these trains are chirped.
In this example, broader spectral analysis (referred to as multiplex CARS, or M-CARS) than that depicted in FIG. 1(a) is achieved through the use of a photonic crystal fiber, a specialty optical fiber with an optical damage threshold that poses severe constraints on the peak power of the pulse train sent through it. As a result, this apparatus achieves a sample flow velocity of less than 200 μm/s—about 50,000 times slower than in conventional flow cytometers and unsuitable for laboratory-based analysis of real samples in a reasonable time. At these speeds, any benefit accruing from broadband CARS spectroscopy is nullified by the resulting impractically low sample analysis throughput.
Existing solutions for analysis of microscopic particles therefore suffer from a number of disadvantages, as follows:
dc impedance, ac impedance, optical extinction, and elastic light scattering in flow cytometry; and brightfield, darkfield, phase contrast, and other traditional optical techniques in microscopy, are not of themselves sufficiently specific to produce information with which to reliably distinguish particles of different types, particularly in many biological applications;
in flow cytometry and microscopy, use of standalone fluorophores provides limited specificity, and additionally requires incubation, which adds costs and reduces efficiency in high-throughput operations;
in flow cytometry and microscopy, use of fluorophore-conjugated monoclonal antibodies requires lengthy incubation, requires complex and expensive measures for shipment and storage of the fragile reagents, and adds significant costs to operations on a per-assay basis;
in microscopy, whether traditional or coupled with advanced spectroscopic techniques like CARS, analysis of stationary samples is much too slow to practically, efficiently, and cost-effectively provide information on a sufficient numbers of particles to confer to such analysis the required statistical significance in most applications;
in CARS spectroscopy as demonstrated to analysis of flowing samples, the current speed of analysis is much too slow for practical implementation, even when trade-offs are made in range of spectral coverage, resulting additionally in very limited compositional information.